skill-creator

Draft, test, and optimize Claude Skill SKILL.md files with quantitative feedback.

3|Updated Mar 22, 2026

Search Tags:#automation #evaluation #iteration #benchmark #skill #prompt

Ready to run

npx skills add https://github.com/eamars/KazusaAIChatbot --skill skill-creator-eamars

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: skill-creator
Source: https://github.com/eamars/KazusaAIChatbot/tree/main/.agents/skills/skill-creator
Command: npx skills add https://github.com/eamars/KazusaAIChatbot --skill skill-creator-eamars

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires yaml.

What problem does it solve?

This Skill provides an end-to-end workflow for creating Claude skills, enabling rapid drafting, evaluation, benchmarking, and iterative improvement in a reproducible manner.

Core Features & Use Cases

Draft new SKILL.md content aligned with triggering criteria and user intents.
Run trigger evaluation loops across train/test prompts, collecting per-prompt results and overall pass rates.
Benchmark different skill configurations and track deltas (e.g., with_skill vs without_skill) to quantify value.
Automatically improve the skill description based on evaluation feedback using the integrated description-improvement tooling.
Save and inspect artifacts (prompts, results, and iterations) to support reproducible skill development.

Quick Start

Write a first SKILL.md for a new skill, then kick off an evaluation loop to measure triggering and iteratively refine the description until it performs well.

Frequently Asked Questions about skill-creator

High-intent search queries and answers about installing and using this skill.

FAQPage Schema

How do I automate prompt evaluation and iteration for Claude skills?▼

Automate prompt evaluation by running trigger evaluation loops across train and test prompts. This collects per-prompt results and overall pass rates to iteratively refine your skill description until triggering performs well.

What is the best way to benchmark different Claude skill configurations?▼

Benchmark different skill configurations by tracking deltas between with_skill and without_skill scenarios. This quantifies value and measures performance improvements based on quantitative feedback from automated testing.

How do I improve skill description optimization based on test results?▼

Improve skill descriptions using integrated description-improvement tooling driven by evaluation feedback. The automated workflow analyzes trigger failures and iteratively optimizes the description to align with user intents.

Can I draft and test Claude skills from inception to production reproducibly?▼

Yes, you can orchestrate drafting, automated testing, benchmarking, and description optimization reproducibly. The workflow saves artifacts like prompts, results, and iterations to support structured skill development from inception to production.

Do I need YAML to write and evaluate SKILL.md content?▼

Yes, YAML is required as a dependency to structure and parse SKILL.md content. It supports the drafting of new skill definitions aligned with triggering criteria and user intents during the evaluation loop.